In this study, the maximum experimental safe gap must be minimized because there are trace levels of Pb in the Yellow River (Gao et al., 2015). All reagents used in this study were guaranteed reagent (GR) grade. The concentrated form of nitric acid (HNO₃) (Fisher^®) used in this study was distilled by an acid purification system (DST-4000, Savillex™). The stock solution of 1 mg L^–1 Pb was prepared in 10% HNO₃ (v/v) by multistage dilution. The 1000 mg L^–1 standard solution was prepared in 500 mL of polytetrafluoroethylene bottles (Asone^®). The Pb standard solution was Cl-forms (+2) and was supplied by the National Center of Analysis and Testing for Nonferrous Metals and Electronic Materials, China. Ultrapure water (>18 MΩ-cm) was obtained by a water purification system (Milli-Q^®).

The labware, including sampling bottles and tubes, had to be made from Teflon materials (Teflon^®) and cleaned according to a strict trace metal cleaning process (Liu et al., 2018). First, a pre-cleaning process was conducted to wash away obvious dirt from the labware, with tap water, 10% cleaning solution (v/v) (Decon90^®), and ultrapure water (>18 MΩ-cm), being used successively. Second, the pre-cleaned labware was soaked in 10% hydrochloric acid (HCL) (v/v) (Fisher^®) and 3% HNO₃ (v/v) for 48 h, respectively. Ultrapure water was used to rinse residual acid off the soaked labware; this was performed five times. Finally, labware were dried in the ultra-clean bench (Class 100) at room temperature (i.e., 20–25°C). After drying, each piece of labware was double-bagged in polyethylene Ziploc (Ziploc^®) bags and stored in a sealed container at a consistent temperature range of 20–25°C.

The study areas were in the Yellow River Delta, which is a typical wetland ecosystem in a warm, temperate zone of the world ( Figure 1 ). Sampling stations were selected according to the locations of cities, villages, farmlands, industrial areas, breeding areas, and salt fields that the seaward rivers passed through during January 2020 ( Figure 1 ). Surface water samples (10–20 cm) were collected and the clean sampling system was used according to the trace metal rule to avoid contamination at all stations. The clean sampling system is composed of a desktop ultra-clean workbench, peristaltic pump, water intake system, syringe filter, and a sampling bottle (Li et al., 2015; Liu et al., 2018). All sample components were connected by C-Flex tubes. The sample process is briefly described below ( Figure 2A ). First, the water intake system was washed five times with ultrapure water before sampling. Second, the pre-washed water intake system was placed 10–20 cm below the surface at the sample water station. Third, the peristaltic pump (Masterflex^®) was turned on so that the water sample passed through the syringe filter (Swinnex^®) with the pre-installed membrane (0.22 μm; Pall^®). The first 10 mL of sample was discarded to prevent interference from the membrane. A total volume of 1 L of the water sample was collected into the fluorinated high-density polyethylene sample bottle (Nalgene^®), after which 1 mL of concentrated HNO₃ was added and the sample stored at room temperature. The other 1 L of water was collected in the fluorinated high-density polyethylene sample bottle and metal was separated by ultrafiltration as soon as possible in the laboratory (<48 h). A total of 70 mL of water sample was taken in a brown glass bottle for DOC and CDOM analysis (<48 h). Further steps were required. The water intake system should be soaked in 3% HNO₃ (v/v) until just before use. The filter membrane should be replaced after the filtration of every sample. The basic hydrological data, such as salinity, pH, and temperature, were measured by the multiparameter water quality analyzer (YSI^®).

The different sizes of colloidal Pb were separated according to centrifugal ultrafiltration (CUF) units (Amicon^® Ultra-15) using a fixed-rotor centrifuge (Cence^®). Four cut-off values (i.e.,<1 kDa,<3 kDa,<10 kDa,<100 kDa) of CUF units were used. The ultrafiltration process is as follows ( Figure 2B ). First, the CUF units were sequentially washed 6–8 times with 0.1% sodium hydroxide (NaOH) (g/g), 0.06% HCl (v/v), and ultrapure water to remove possible contaminants (Lu et al., 2020). Second, 15 mL of sample was loaded to the obtain the different cut-off values and centrifuged for 50–70 min at 4800×g in the rotor centrifuge. Approximately 0.5 mL of ultrafiltrate was obtained from the inner ultrafiltration tube and transferred to the pre-cleaned 10-mL polypropylene sample centrifuge tube (Falcon™). After CUF, 1 mL of 3% HNO₃ (v/v) was added to the inner tube and soaked for 2 h to extract the residual colloidal metal. The extraction was performed three times and mixed in the sample tube. The volume of colloidal sample was adjusted to 5 mL with 3% HNO₃ (v/v). Before measuring, colloidal Pb samples were stored in an ultra-clean bench (Class 100; at 25°C) for more than 1 week to ensure that the colloidal organic matter was fully decomposed. Finally, 0.2 mL of the decomposed solution was diluted to 10 mL with 3% HNO₃ (v/v); this solution was then used for detecting colloidal Pb.

A method blank and solvent blank of the ultrafiltration were used simultaneously with samples to assess errors. It should be noted that the actual cut-off values of the CUF used in this experiment were generally higher than those stated by the manufacturer (Xu et al., 2018; Lu et al., 2020). This can usually lead to biased perceptions of the colloidal trace element concentration in the ultrafiltrate by researchers. We used standard macromolecules, including vitamin B₁₂ and standard fluorescent-tagged dextrans, to calibrate the CUF units according to previous studies (Lu et al., 2019; Lu et al., 2020).

The fractionation factor (F) was defined as the ratio of the concentration of truly dissolved Pb and colloidal Pb to evaluate the fractionation behavior of Pb by the colloids present in the system (Liu et al., 2019):

where C_Pb3,c and C_Pb3,t are the concentrations of Pb in colloids, respectively. F_c/t is the fractionation factor between colloidal Pb and truly dissolved Pb. When the F_c/t > 1 or< 1, the Pb preferentially migrated to colloidal matter or the truly dissolved phase.

The UV absorption spectrum of CDOM was measured using a UV–Vis spectrophotometer (TU-1950, Persee^®). Ultrapure water was used as a blank and for baseline scanning. The samples were spectrally scanned in the range of 200–800 nm, and the scanning interval was 1 nm. All absorbance values were deducted by the mean value of the 680–700 nm range to eliminate the refractive index differences and baseline drift.

The absorption coefficient a(355) was selected as the parameter represent the concentration of CDOM and was calculated from the following equation (Peuravuori and Pihlaja, 1997; Wu et al., 2022):

where a(λ) (m^–1) is the absorption spectrum at the wavelength of λ nm; A(λ) is the absorbance at the wavelength of λ nm; and L is the length of cuvette.

The specific ultraviolet absorbance (SUVA₂₅₄) was defined as the ratio of the absorption coefficient at the wavelength of 254 nm to the concentration of dissolved organic matter (DOC) of the sample. This parameter could represent the aromaticity of CDOM in the aquatic system. It was calculated from the following equation (Peuravuori and Pihlaja, 1997; Wu et al., 2022):

where SUVA₂₅₄ (l (mg m)^–1) is the specific ultraviolet absorbance; a(254) is the absorption spectrum at the wavelength of 254 nm; and C_DOC is the concentration of dissolved organic matter.

The ratio of the absorption coefficient (E2 : E3) at a specific wavelength was used to represent the change of the relative molecular weight of CDOM and was calculated using the following equation (Peuravuori and Pihlaja, 1997; Wu et al., 2022):

where E₂/E₃ is the ratio of the absorption coefficient of the water sample at the wavelength of 250 and 365 nm; a(250) is the absorption coefficient at the wavelength of 250 nm; and a(365) is the absorption coefficient at the wavelength of 365 nm.

In this study, the concentration of different sized colloidal Pb was determined by using an inductively coupled plasma mass spectrometer (ICP-MS, iCAP RQ, Thermo Fisher Scientific Inc). The internal standard [indium (In₄₉), 10.00 μg^–1] was spiked in every sample. To maintain the stability of the instrument, we used 1.00 μg L^–1 tune solution [lithium (Li₃), cerium (Ce₅₈), barium (Ba₅₆), In₄₉, and uranium (U₉₂)] to monitor the oxide formation (BaO/Ba< 3%). The isobaric interference on Pb was corrected by monitoring ²⁰⁰Hg and ¹¹¹Cd (Cd₄₈). The data of ¹⁰⁸Pb, ⁶⁵Cu (Cu₂₉), ⁶⁸Zn (Zn₃₀), ⁸⁵Sr (Sr₃₈), ⁸⁹Y (Y₃₉), and ⁹⁵Mo (Mo₄₂) were used to monitor the potential polyatomic interference. ICP-MS analysis was conducted with 99.999% argon at the optimized plasma gas flow rate (GFR) of 15.0 L min⁻¹, an auxiliary GFR of 1.2 L min⁻¹ and at a nebulizer GFR of 0.88 L min⁻¹.

To ensure the accuracy of the method, we measured the method blanks to verify and control the reliability of the analysis data. The method blanks were determined by using an ultrapure water sample. The method blank followed the same process as sample preparation as follows: 3% HNO₃ extraction, 3% HNO₃ dilution, and ICP-MS parallel determination (n = 11) (Lu et al., 2020). The quantification limits [i.e., limit of quantification (LOQ)] for Pb were 10 times that of the standard deviation (σ) of the blanks, and the detection limits [i.e., limit of detection (LOD)] were 3σ ( Table 1 ). To verify LOQ and LOD, we used the GBW08608 and GSB07–1183 reference materials to test the linear correlations (R ²) of the method. The linear ranges were 0.2 μg L^–1, 0.5 μg L^–1, 1 μg L^–1, and 2 μg L^–1. The performance was shown in Table 1 . 0.50 μg L^–1 of Pb standard sample was determined every 12 samples to measure the stability of the ICP-MS.

ICP–MS, inductively coupled plasma mass spectrometer; LOD, limit of detection; LOQ, limit of quantification.

The fractionation method, solvent used, and detection error will determine the result of Pb in the different samples. We used the mass balance (M) to monitor the precision of the result. The mass balance was defined as the ratio between the dissolved concentration of Pb of the non-fractionated sample and the summary of the different fractionation concentrations of Pb. Once the difference value between M and 100% was less than 10%, we believe that there on loss during the experiment. Moreover, once the value reached greater than 10%, we believed that the loss of Pb affects the result of the experiment (Liu and Gao, 2019). The mass balance ratio results of the different stations are shown in Figure 3 . The mass balance ratio difference between M and 100% was 9%–10%. This means that the level of error introduced by the method, solvent and detection was overlooked in this study. Notably, the recovery ratio varied with the location of the sample station. Higher recovery ratios were observed in stations influenced by human activities. It is shown that there is an inference relationship between human activities and Pb. Intense human activity often leads to a stronger release of Pb into the environment (Zang et al., 2020). It will lead to the introduction of Pb contamination during field sampling or pre-treatment, ultimately causing high recovery rates (Chen et al., 2022).

The actual cut-off value for CUF was consistent with previous work because we used the same batch of products in this study (Liu et al., 2018; Lu et al., 2019; Lu et al., 2020). The actual MWCO of the 1, 3, 10 and 100 kDa CUF units were estimated to be 2, 7, 32 and 393 kDa, respectively. This indicated an overestimation of the colloidal trace element concentration in the ultrafiltrate of this study. However, the ultrafiltration performance of the CUF could still precisely fractionate the Pb into different sized colloids. Therefore, the experimental results could reflect the migration behavior of Pb from truly dissolved to colloidal phases in rivers flowing into sea.

Hydrographic parameters such as water temperature, salinity and pH of the rivers are described in Table 2 . Water temperature of all the stations fluctuated in the range of 1.6°C to 7.8°C, with gradual decreases toward the direction of input sea. In addition, more than 80% stations were lower than 5°C, which means that the biogeochemical effects between microorganisms and the metals were very small and could be negligible (Cao et al., 2018; Marcinek et al., 2022). A notable phenomenon was that the water temperature of the Zihe-Xiaoqinghe river system was significantly higher (i.e., ≈1°C) than that of other rivers in this study. Higher water temperatures (i.e., >5°C) were recorded at the station near the village and the highway. The pH of most rivers was above 8.10, and the highest value appeared at the station of Shenxiangou River, with a value of about 9.25 ( Table 2 ). There is an obvious high pH (>9.00) area in the Guanglihe River section passing through the urban area of Dongying City ( Figure 3 ). This higher pH may be related to the weak flow of the river and the salinization of the land. To prevent sea tide intrusion, the flow of the Guangli River into the sea is controlled by a dam with fixed opening and closing times. The damming of the river by humans leads to a weaker flow of the Guangli River. A weak Guanglihe River flow is susceptible to saline seepage from salinized sediment, which often increases the pH of the water (Xie et al., 2019). In addition, in the reach of the Zihe River (which merges into the Xiaoqinghe River), there are stations with pH lower than 8.00, with values around 7.70–7.90. The station with the lowest pH (≈4.18) appears in the Zhimaihe River section, close to an industrial area, and the low pH river section continues for about 15 kilometers. From the potential results, all rivers are reducing environments under alkaline conditions, with the river pH >7.00 ( Table 2 ). Significant differences were found in the salinity of the rivers. For the main rivers, such as Yellow River and Xiaoqinghe River, changes of salinity were not detected owing to the impact of the strong supply of upstream fresh water. There is no significant change of the salinity at the stations of the Yellow River, which remained at 0.42. In the Xiaoqing River, whose hydrodynamic force is weaker than that of the Yellow River, the salinity only reaches 7.00 at the station near the estuary and did not exceed 3.00 at other stations. For the trunk channels, the salinity changed drastically. A pronounced increase of salinity was found along the direction of the seaward river output into the sea. Some sample stations with a salinity greater than 20.00 appeared in the Xiaodaohe River and Shenxiangou River. The salinity changes observed in the Guanglihe River and Yongfenghe River were relatively slow, but the salinity exceeds 15.00 at the stations near the estuary.

The seaward rivers could be classified into three categories according to the discussion in Section 3.1. The Xiaoqinghe River and the Yellow River are natural rivers (NR). Trunk canals of the Yellow River are divided into two categories according to the influence of tidal seawater. The Yongfenghe River and Xiaodaohe River are rivers strongly influenced by seawater (SFR), whereas the Zhimaihe River, Guanglihe River, and Shenxiangou River are weakly influenced by seawater (WFR).

In this study, the migration behavior of Pb in the Yellow River was used as a reference location. This sample reach is not affected by industrial and domestic wastewater outfalls and agricultural activities in the winter. In addition, the concentration parameters showed that colloidal Pb (1.27 ± 0.06 μg L^–1) is basically stable in the Yellow River. This means that the Pb migration mainly occurs between the truly dissolved and particulate or sediment phase related to the weakened hydrodynamics. As the Yellow River flows into the sea, the main channel gradually becomes wider with a reduction in flow (Du et al., 2022). Low hydrodynamic action increases the chance of truly dissolved Pb entering the sediment with particle settlement (Moore et al., 1996). However, anthropogenic activities significantly influenced the Pb concentration behavior regarding partition patterns between colloid and truly dissolved Pb in other rivers. The change trend of the fractionation factor (F) of Pb in the different phases is shown in Figure 6 . The Xiaoqing River sampling area consisted of three reaches, the upper reach (SX1), the middle reach (SX2→SX6) and the estuary (SX7→SX8) ( Figure 1 ). The upper reach and the estuary are greatly affected by chemical industry, salt industry, and port activities, whereas the middle reach is less affected by human activities because there is little agricultural activity in winter. Sewage discharge from the refining and chemical industry likely led to the surge of colloidal Pb (to 19.60 μg L^–1) at the SX1 station. Downstream in SX2, the colloidal Pb decreased sharply and truly dissolved Pb was stable. This phenomenon might be due to the unstable upstream chemical source of colloidal Pb, and input sources disappeared in this region (Teien et al., 2004). In addition, the colloidal Pb and truly dissolved Pb in the Xiaoqing River increased significantly after passing through some large residential areas (SX4). This phenomenon is related to the direct discharge of sewage from this area (Gu et al., 2022). Domestic wastewater produced by residential areas often contains large amounts of Pb and organic matter (Deng et al., 2020; Silva et al., 2020). In addition, this residential area is also home to petrochemical factories. The factories could produce wastewater containing high levels of Pb and organic matter (Wang et al., 2017; Dong et al., 2022). Hence, the domestic mixed factory discharge is likely to be an important source and disturbance factor of Pb in the Xiaoqing River. Notably, total dissolved Pb changed little after flowing through the area with less human influence (SX4→SX6). Furthermore, the dissolved Pb mainly migrated from the colloid phase to the truly dissolved phase ( Figures 5 , 6 ). This phenomenon is the opposite of the behavior of dissolved Pb in the Yellow River. It may be related to a difference of types of colloidal organic matter that are present in the Yellow River and the Xiaoqing River (Lu et al., 2021). An increase in colloidal Pb was observed after passing through the Yangkou port with its intensive human activities (SX8). The significant reduction of truly dissolved Pb in this region may be related to the rapid increase in salinity. Similar phenomena as were observed in the Xiaoqing River was also observed in the Zhimaihe River, the Yongfenghe River, and the Xiaodaohe River. Both the concentration and the fractionation of colloidal Pb significantly increased after those rivers passed through ports or residential areas ( Figure 5 ). However, the migration of Pb was more complex in the Guanglihe River and Shenxiangou River because of the lower hydrodynamic force combined with the higher flow through urban areas. Previous studies have shown higher levels of Pb and more variable biogeochemical parameters, such as DOC and CDOM, in rivers flowing through urban areas (Liang et al., 2018; Liu et al., 2022). Overall, domestic and industrial sewage were the dominant factor regarding partitioning between truly dissolved Pb and colloidal Pb in the rivers flowing into the sea in the Yellow River Delta in the winter.

As shown in Figure 7 , total dissolved Pb sharply decreased with salinity gradient, expect for the Zhimaihe River. It previously demonstrated that salinity could induce the flocculation of Pb in river water and this behavior was non-conservative (Wu et al., 2015). However, Pb in different phase express different behavior. Truly dissolved Pb gradually decreased with the salinity gradient and expresses a non-conservative behavior ( Figure 7 ). But for colloidal Pb, both non-conservative and conservative behavior can be observed. The removal behavior of Pb mainly occurs in the truly dissolved state. This phenomenon may be the result of the combined effect of the speciation of Pb ions and the flocculation effect of salinity on low molecular organic matter (Xu et al., 2020; Nghiem et al., 2022). At elevated pH (>8.00), Pb ions easily transform to PbCO₃ precipitate with increasing of salinity (Powell et al., 2009). Some studies had demonstrated that organic Pb chelates was the dominate species in the truly dissolved Pb in natural water (Raudina et al., 2021). Therefore, the gradual increase in salinity and higher pH promotes the precipitation of truly dissolved Pb ( Table 2 ). To clarify this combined effect, we performed principal component analysis of the pH and salinity variables with different state Pb and shown in Figure 8 . From the figure we could found, pH values and salinity of the river water have similar negative correction with the dissolved Pb, especially for the truly dissolved Pb. But for colloidal Pb, pH expresses a positive correlation ( Figure 8B ). Meanwhile, pH had more intense effect than salinity for all state Pb ( Figure 8 ). This behavior may dominate by the interaction between salinity, pH, and the accumulation of colloid matter (Cantwell and Burgess, 2001). Salt may induce organic matter to aggregate with each other for flocculation (Lasareva et al., 2019), but not completely flocculate the large sized colloidal matter (Herzog et al., 2019). The colloidal material can be stable under high pH and high salinity such as humic acid (Liu and Gao, 2019; Furukawa et al., 2014). The coagulation of inorganic colloids would reduce while above a certain salinity range, especially in presence of organic matter (Wang et al., 2015; Zhang et al., 2015). Therefore, temperature might another factor to affect the migration of Pb. It had demonstrated that temperature tended maintain the stabilization COC in the winter river (Gong et al., 2022). The principal component analysis between the temperature with different dissolved state Pb also performance ( Figure 8 ). Low temperature had a positive correction with the state of dissolved Pb ( Figure 8 ). It means temperature might an important factor to maintain the dissolve state of Pb. However, pH had a significant negative correction with the truly dissolved Pb ( Figure 8C ). It means truly dissolved Pb removal from the water in the estuary area where the salinity increasing rapidly, higher pH values, and especially in the low temperature. These results suggested that low temperature, high pH, and increased salinity have a combined or together effect on the removal of Pb form river water. These three biogeochemical factors associated reduced 33.3%≈95.4% of Pb entering to the sea with the seaward river.

Colloidal Pb was stable, and the colloidal organic carbon (COC, represented by the concentration of DOC) decreased with the direction toward the sea in the Yellow River ( Figure S1 ). However, 10–100kDa colloidal Pb demonstrates a significant correlation with the change of COC ( Table S1 ; Figure S1 ). This phenomenon might be related to the components of colloidal matter. It has demonstrated that different colloidal components could strongly bind the Pb₃ ions in water, and this binding ability varies with the character of the colloidal matter (Stolpe and Hassellov, 2010; Feng et al., 2022). However, it is difficult to clarify this effect because of the lack of analysis techniques that exactly separate and quantify colloidal matter components (Klun et al., 2019). Fortunately, some studies have discovered a significant relationship between CDOM of colloidal matter and Pb ions (Javed et al., 2017). Hence, we discussed the correlation between the molecular weight and components of CDOM and the colloidal Pb to explore this effect (Stolpe and Hassellov, 2010). The CDOM-UV parameter in different stations was shown in Table 3 . We used the absorption constant at 355 nm to represent the concentration of CDOM, SUVA₂₅₄ represents the aromaticity of CDOM, and a250/a365 represents the molecular of CDOM.

In the Yellow River, the concentration of CDOM and COC was decreased but SUVA₂₅₄ was enhanced. Meanwhile, the a250/a365 value (10.09–14.45) showed that the molecular weight of CDOM was higher ( Table 2 ). However, the concentration and the aromaticity of CDOM had a poor correlation with the different sizes of colloidal Pb ( Figure S2 ). This shows that non-aromatic organic matter exhibited obvious aggregation or elimination, and inert soluble organic matter increased in concentration (Lu et al., 2019). In addition, Pb ions tend to partition in high-molecular-weight colloidal organic matter ( Figure 5 ). Hence, the stable colloidal Pb in the Yellow River indicates a likely close relationship with this metal to the high-molecular-weight aromatic-like matter. Therefore, results suggest that the migration of colloidal Pb may be related to aromatic CDOM in areas with less disturbance from human activities and freshwater areas. In the Xiaoqing River, which is also a natural river, the changes of CDOM and colloidal Pb were more complicated. Except for the Zihe input station (SX4), the aromaticity and molecular weight of CDOM in the Xiaoqing River were lower ( Table 2 and Figure S3 ). This illustrates that the distribution of Pb₃ in different sizes of colloidal matter have a close relationship with the high-molecular-weight aromatic matter. However, in the trunk channels which are greatly affected by the tide, the salinity induced the aromatic components of CDOM to gradually aggregate and promote agglomeration of colloidal Pb. Meanwhile, the trend in change of the CDOM abundance was conformation for all sizes of colloidal Pb ( Figures S4–S8 ). Human activity mainly causes input of low-molecular-weight CDOM, which is also an important migration destination of Pb ions ( Figure 5 ). The CDOMs are not stable and have relative ease of decomposition, inducing a complex distribution of Pb in different sizes of colloidal matter ( Table 2 ). Overall, the highest concentration of colloidal Pb₃ was observed in the station that has stronger aromaticity of present CDOM ( Table 2 and Figure 5 ). Furthermore, Pb was also more inclined to be found in the large-sized colloidal matter. However, in winter, the distribution of Pb in large-sized colloids is relatively stable because of the low decomposition effect of the low temperature.

To further verify those findings, we compared the fractionation behavior between Pb in the truly dissolved phase and colloidal organic matter phase in other rivers into the sea. Table 4 shows the concentration of Pb in the truly dissolved phase and colloidal phase in different areas. It can be observed from Table 4 that the truly dissolved Pb was the main components in the freshwater (Javed et al., 2017; Gandois et al., 2020; Lu et al., 2020; Zhang et al., 2022). However, with the enhancement of salinity, the colloidal Pb became the main components of the dissolved Pb in the brackish water environment (Illuminati et al., 2019; Lu et al., 2020; Pavoni et al., 2020; Feng et al., 2022; Nasrabadi et al., 2022). Many studies have demonstrated that Pb is prone to sedimentation in estuary where salt water and freshwater are mixed (Pavoni et al, 2020; Nasrabadi et al., 2022). This phenomenon of migration distribution is more pronounced in the estuarine area. For example, colloidal Pb exceeded 50% of the total dissolved Pb in most estuarine areas (Illuminati et al., 2019; Lu et al., 2020; Pavoni et al., 2020; Feng et al., 2022; Nasrabadi et al., 2022). In the Krka River estuary, this fractionation behavior was close to 100% (Nasrabadi et al., 2022). Combined with the phenomenon of this study, the sedimentation of Pb mainly occurred in the small molecule organic phase ( Figure 5 ). In addition, the increase in salinity promoted the migration of Pb to macromolecular colloids ( Figures 5 , 6 ). This suggests that macromolecular colloids matter will be the main dissolution state of Pb from rivers to the ocean.

From the concentration change of Pb observed in this study, we could find that, excluding the influence of industrial pollution input, most Pb was precipitated in the estuary ( Figure 7 ). This indicates that the ecological risk of Pb in the river water was relatively low. However, the disturbance by anthropogenic factors increases the risk of Pb migrating to the sea (Wang et al., 2022). As the winter passes, the strong hydrodynamic force caused by Yellow River flows will induce scouring of the trunk channel riverbed (Mistri et al., 2019). Pb deposited in surface sediments in winter will be carried into the sea by later floods. Furthermore, the water body replacement of the urban water system and river dredging works will periodically bring urban sedimentary Pb into the offshore environments. Meanwhile, these rivers do not have a strong hydrodynamic force and vast estuary areas like the natural Xiaoqing River and the Yellow River. Hence, trunk canals (Yongfenghe River and Xiaodaohe River) and urban water systems (Guanglihe River and Shenxiangou River) may be more likely to transport sedimentary Pb to the sea.

There will still a large amount of Pb transported into the sea with the river in the Yellow River Delta. Free Pb ions were easily precipitated in the water with multiple geochemical factors. However, the colloidal Pb after salinity selection at the estuary was more stable in seawater (Chambari et al., 2018). Colloidal Pb is also more easily taken up by organisms in the water because of its organic characteristics. The ecological risk of colloidal Pb was higher than Pb ions because of the easier enrichment of this speciation in the food chain (Lintner et al., 2019). In addition, mariculture, ports, and coastal factories have become the main sources of colloidal Pb. The wastewater discharged from these areas has high salinity. Colloidal Pb in that wastewater will go through a long time of salinity selection. For this, the sedimentation of colloidal Pb will become reduced at the estuary (Lee et al., 2014). A large amount of active Pb will directly input into the seawater, increasing ecological risks. Hence, the pollution by mariculture, ports, and coastal factories should require further attention to reduce ecological impacts.